This work investigates procedures for testing equivalence and estimating the ratio of marginal probabilities for paired dichotomous variables. A procedure is developed for conducting a test of equivalence and finding a confidence interval for the ratio of probabilities. This is appropriate when there is no gold standard of disease. A 2 d.f. chi-squared test is developed for the case in which a gold standard exists. These are both extensions of McNemar's test for paired dichotomous data.